KG-Eval Report

📈 Performance Overview

🔢

Scale & Richness

Measures the breadth and depth of extracted information

Entity Count

4,159

Total number of unique entities identified

Relationship Count

6,916

Total number of relationships extracted

Property Fill Rate

100.0%

Excellent Completeness of entity properties

Relationship Types

3862

Diversity of relationship types used

🕸️

Structural Integrity

Evaluates graph connectivity and topological health

Graph Density

1.6629

Excellent Ratio of actual edges to possible edges

Largest Connected Component

69.9%

Good Percentage of nodes in main connected component

Connected Components

1189

Number of disconnected graph components

Singleton Nodes

27.5%

Percentage of isolated nodes (lower is better)

✅

Semantic Quality

Assesses accuracy, consistency, and relevance of extracted knowledge

Entity Normalization Score

74.2%

Good Consistency in entity naming and representation

Factual Precision

🤖 92.0%

LLM-evaluated factual accuracy of extracted information

Potential Alias Pairs

1073

Entity names that might refer to the same concept

Contextual Relevance

🤖 26.0%

LLM-evaluated relevance to source context

⚡

Efficiency

Measures effectiveness of text-to-knowledge transformation

Knowledge Density

17.98

Excellent Average knowledge elements per text chunk

Productive Source Ratio

100.0%

Excellent Percentage of source texts that contributed to KG

Average Text Length

1085

Average length of source text chunks

Source Coverage

0.0%

Percentage of source content linked to KG elements

💡 Recommendations for Improvement